Author Topic: Batch learning?  (Read 2946 times)

jz1977

  • Newbie
  • *
  • Posts: 24
    • View Profile
    • http://dageeks.com
Batch learning?
« on: April 11, 2004, 10:48:36 pm »
hello,

I have bought Ultra Hal Assistant version 5.0.57, but my main question is, I have over 15,800 text files that are 65k in size i want hal to learn.  But i have not found a way to have hal learn in batch mode.  Is there a way or a tool i can use to import all 15,800+ text files so i dont have to do one by one?  If anyone knows of an way i can do this, i would much appreciate it.  My other concern is, with a database this big, will the response be slow,  my guess it will be.  But with my hard work, I beleive i will have the smartest bot known to exist, that knows everything humanly possible from, makeing you feel better when your depressed ->to-> teaching you how to configure all cisco Routers. By the time i am done upgrading hal, he will have over 1 GIG of a brain's database.  Let me know what i should do in this situation.

Thanks,

-Jeff
If you truly beleive it, without any doubt, then it must be True and real.

vonsmith

  • Hero Member
  • *****
  • Posts: 602
    • View Profile
Batch learning?
« Reply #1 on: April 12, 2004, 01:37:57 pm »
jz1977,
Hal's learn from text file mode is very primitive. There is a file size limit on that type of learned information that will limit the knowledge files to about 2MB. The learned information won't really be stored in a way that is clearly accessible to Hal either. Several people here on the forum, including myself, have been trying to write programs or VBscript files to add a real "learn from text file" capability. I don't think anyone has succeeded yet. It is a challenging problem, but not impossible to solve.

I don't know how well Hal will operate with 1GB of data. In theory it should work. I can't be certain if you might hit a file limit or something. Perhaps others on the forum might be able to comment.


=vonsmith=
 

spydaz

  • Hero Member
  • *****
  • Posts: 670
    • View Profile
    • http://www.spydazweb.co.uk/
Batch learning?
« Reply #2 on: April 17, 2004, 10:21:47 am »
I HAVE FOUND THAT THE BEST WAY TO A DATA TO A HAL BRAIN
IS TO GET MY TEXT FILES AND EDIT THEM IN TO STATEMENTS EITHER TRUE OR FALSE. AND CREATE AN INDIVIDUAL TOPIC

A SAMPLE TOPIC IS SHOWN IN THE FORUM BELOW

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=1224

THE SCRIPT IS EASY ENOUGH TO EDIT FOR EACH TOPIC CREATED.

REMEMBER HAL WORKS ON KIND OF A keyWORD TRIGGER SYSTEM
if A KEYWORD IS FOUND. then THIS WILL TRIGGER A RESPONSE FROM ONE OF HIS STATEMENT*.BRN FILES

ALSO HAL'S Q&A FEATURE (TOPIC BRAIN) IS ALSO A KEY PART OF HAL'S MAIN FUNCTION. BY USING THE BUILT-IN TOPIC FUNCTION (Q&A) YOU CAN GET STATIC ANSWER TO STATIC QUESTION TOO.

WHEN HAL learns from a text file (IT'S NOT VERY GOOD) IT DOSENT Guarentee THAT YOU COULD EVER ASK HAL ANY THING ABOUT THAT FILE AND GET A RESPONSE.

i think that the best plan would be to WRITE A SCRIPT that learns from DATABASE files. or (COMMA,SEPERATED,TEXT,FILES) THIS WOULD GREATLY ENHANCE HAL'S LEARNING CAPABILITY.

WHAT COULD AN ARTIFICIAL INTELIGENCE LEARN FROM A TEXT FILE THAT COULD HELP YOU (conversation)

word trigger response is a bit like a query>>>>> database type <<<<<


HMMM i really dont know>>>>>>>

MAYBE THE CREATOR COULD THINK SERIOUSLY ABOUT THESE CONCEPTS.....


I BELEIVE THAT THE army&govenment ALREADY HAVE A SIMULAR SYSTEM WHICH USES THESE INFLUENCES......

THIS MAY ONLY BE FOR MEMBERS OF THE PUBLIC (and the best on the market) BUT ........ ........ .... OPPs I've said too much


thanks all